The construction of the data platform can be designed as a product. In a broad sense, all data-driven products can be called data products (such as data reporting platform, DMP, BI platform), and in a narrow sense, it is the data platform within the company. What we are going to discuss today is mainly to build a data platform within the company.
The company's internal data platform mainly provides data processing, analysis and display for each business, and is used by all internal departments, involving data collection (filling in), data integration, business report production, salesperson data analysis and visualization data The purpose is to enable all businesses within the company to drive and assist decision-making through data. To put it simply, it is to drive the internal data operation of the company through the data analysis platform.
So how to design a data analysis platform that conforms to the reality of the enterprise and can solve practical problems?
1. The leader of platform construction needs to have a deep understanding of each piece of business needs, and know what kind of data each business department wants to see and what kind of analysis reports they need; whether these data can be obtained now and whether they need to be collected; Through these data analysis, how does the business department promote and improve the business, and whether it has the meaning of increasing value.
2. The design of the platform needs to design the content that meets the needs of users according to the requirements of the business, and the product must have a hierarchy and structure. If a data report is designed to meet the needs of both management and front-line business personnel, the experience of such a data product is likely to be poor. Because leaders and business people have different concerns, they have different perspectives on data. Leaders often need some key indicators that can help grasp the general direction, and want to know what is the problem behind these indicators? what is the reason? Therefore, reports designed for leaders need to be intuitive and easy to understand, and can locate problems based on these indicators. The business personnel are more concerned about the execution of the business, and the data they pay attention to are often very fine-grained, and they need to know the details of each indicator.
Reports of Leaders' Concern (made by FineReport)
A breakdown of metrics that sales leaders care about (by FineReport)
3. The data platform must pay attention to data quality, standardization and unification. Because the data analysis platform is for all businesses, it is particularly difficult to ensure that all departments of the company have a consistent understanding of the data. (For example, the "broken code" in the clothing industry, from the perspective of the leadership, if all the sizes of the clothing in the company's warehouse are incomplete, the code is broken; from the perspective of the warehouse keeper, if the clothing size in the warehouse is incomplete, the code is broken; from From the point of view of the salesperson of the store, if the size required by the customer is currently out of stock in the store, the code is broken). Companies have different business systems, database distribution, different data calibers, and data silos, resulting in different data quality and structure. The more this is the case, the more important the data construction of the data warehouse is. The data quality of the platform depends on the underlying data model of the data warehouse, so a good data warehouse largely determines the data quality of the data analysis platform.
4. In terms of tool selection, there are reporting platforms and BI. The reporting platform is suitable for building a basic standardized data analysis platform, from detailed reports (tables), project files, document reporting, data reporting, data reporting, business theme analysis, all demos in this article are made with FineReport, focusing on display and report management. BI focuses on analysis. When you get data, you can drag and drop dimensions for analysis. Unlike reports, which are limited by template frameworks, they cover simple detailed reports, analytical reports, and topic analysis. The production is much simpler, and the processing performance of large data volumes is also strong. Many, on behalf of: FineBI, Tableau.
Next, we will explain it from a practical point of view.
A company is an e-commerce company, what data do the various departments of the company need to see? First, collect the data indicators that are commonly used in daily life, which ones are frequently queried and which ones need to be filled in daily. These have been filed in the past operations and are easy to collect. This constitutes the daily basic query report, which accounts for the most.
Secondly, we can observe what are the KPIs of each department? What are the KPIs delegated to the team as well as to the individual? Can it be quantified with specific numbers? If you don’t understand the KPIs of the department responsible for support, seek help from the department leaders (users know what they want best), how can you design a good data report without implementing the indicators? Usually rough, for example, the KPIs of the purchasing department are basically sales, the number of orders, gross sales profit, and procurement costs; the KPIs of the operation department are the description of new and old users, retention rate, repurchase rate, user churn, conversion rate, and marketing department. KPI is PV/UV, number of new customers, which constitute the daily management report of each department.
After knowing the core KPIs of each department, the next step is to report on some specific scenarios. Thematic reports often record an event, and the indicators in it are all related to each other.
For example, I would like to know how my performance is doing this month? Does the team rank up or down? What is the breakdown of performance results for each individual? - This constitutes the daily assessment report.
For another example, when analyzing products, it is necessary to focus on the relative market share and market growth rate of certain products, and a Boston matrix analysis should be established in order to effectively allocate resources to a reasonable product structure (of course, there are other analysis angles)— - This is category analysis.
Product Analysis (made by FineReport)
Let’s start with business analysis and usage scenarios. Take Xiao Wang from the procurement department as an example. How does he view data?
Every morning, I want to know how the sales were yesterday? Therefore, a basic query report should be designed for him at this time. This data report should have the following functions and contents:
1. Be able to view the data of yesterday and past periods; be able to make statistics by product and region; be able to compare with the past average to see if there is a big change in the trend of a certain product in a certain region, whether it is advertising, The promotion of the event has little impact. Secondly, what is the data of each month and each quarter, which can be displayed automatically in the selected time period.
2. The richer the indicators, the better. If the sales volume drops, check whether the number of orders has dropped, and the number of orders has not changed. Is it a problem with the unit price of the customer, whether it is due to the product structure or the adjustment of the activity threshold.
3. The finer the granularity of data analysis, the better. For example, the data granularity can be drilled down from the country to the province, and from the province to the city, so that when the transaction volume drops, you can know which province and which city has the problem? targeted solutions.
After the above steps, the prototype of the platform can be formed and the planning document can be formed. But a data analysis platform, no matter how richly planned in the early stage, cannot be achieved overnight. The company's business is constantly changing, and the content of the analysis is getting richer and richer. In the process, it always needs to be constantly adjusted and adjusted. Only after continuous improvement can a more tailored and practical platform be formed.